Resource constraints can invalidate an elegant network

A precedence-feasible programme can still be impossible to resource. The network may allow equipment inspection and receiving acceptance to occur at the same time, while both require the same qualified specialist. The app calculates no resource allocation and cannot discover this conflict. Its early dates describe what the topology permits under unconstrained start availability. A delivery manager should therefore treat the result as an input to resource planning, not as proof that the organisation can execute all permitted activities simultaneously.

Resource constraints include people, equipment, workfaces, specialist review capacity and operating access. Their significance depends on timing and substitution. A common labour category may be replenishable; a certified commissioning authority may be scarce and non-substitutable without qualification. A resource record should specify what capability is required, when it is needed, how availability is evidenced and what alternatives are credible. Headcount alone is a weak basis if the work needs a particular competence, permission or location.

An overlap can be physically feasible but organisationally constrained. The receiving works and equipment supply branches may use separate execution teams yet rely on the same design authority to resolve interface questions. The apparent parallel network then concentrates review demand. If design clarifications arrive together, queues can develop outside the model. Adding a generic contingency duration to both branches may obscure the shared cause. A resource and decision-capacity review should identify the common dependency and consider whether planned sequencing, additional qualified capacity or better release completeness addresses it.

Resource allocation can create a new critical sequence. If a specialist must finish one package before starting another, the operational programme may need a resource-driven relation or a separate levelling method. The typed dependency register should mark the cause so future changes remain understandable. A different resource arrangement could remove that sequence. If the relation is presented as physical logic, reviewers may overlook this option. If the relation is removed without obtaining alternative capacity, the programme becomes attractive but infeasible.

The model boundary should be repeated when its result enters a decision. A sponsor comparing the reference and duration-switch studies can see the conditional effect of shorter supply assumptions. The sponsor cannot conclude that adding labour will achieve those assumptions. That requires a separate productivity and resource basis, including supervision, access, fatigue, quality and cost consequences. Resource mobilisation is an intervention with evidence requirements, not a numerical adjustment justified merely because the network benefits from it.

A resource conflict should be described by the accepted outcomes at risk. Saying that a specialist is overloaded is a start, but the delivery decision needs to know which release events compete and what happens if either waits. One event may have float in the conditional network; the other may govern accepted completion. Even then, assigning the specialist to the critical event is not automatically correct. The supposedly flexible task may have a fixed review window or a consequence outside the app's perimeter. Compare the complete obligations before choosing a sequence.

Resource availability evidence should have a validity period. A supplier's indicative promise of a crew differs from an executed mobilisation arrangement. A named person may be available at the planned date but committed elsewhere if the project moves. The schedule can therefore lose resource feasibility after a duration change even when the network's task means remain credible. A revised forecast should trigger a check of resource reservations and external commitments, especially near commissioning or shutdown windows where alternatives may be limited.

Resource substitution requires an acceptance basis. A technically capable substitute may lack authorisation for a regulated test or familiarity with the configuration. The record should name the qualification, induction or approval needed before substitution is executable. Treating every person within a job category as interchangeable can overstate recovery options. Conversely, assuming only one person can act without checking permitted delegation can understate them. The control question is what evidence makes the substitute valid for the specific obligation.

A resource intervention may transfer rather than remove risk. Adding another contractor can reduce a queue while increasing interface and supervision demands. Moving work to another shift can accelerate elapsed progress while affecting quality inspection or fatigue controls. Splitting packages can make parallel execution possible while creating new acceptance boundaries. These consequences should appear in the recovery record. The small app can test a new duration or topology scenario after the intervention is defined, but it does not value the resource arrangement or its wider exposures.

A resource plan should preserve both demand and capacity assumptions. A simple list of assigned names does not explain how much attention tasks require or whether the assignments overlap. The integrated manager needs enough information to see material conflicts without reproducing every local timesheet. Package managers can maintain detailed allocations; the integrated resource view should focus on constrained capabilities and release windows that govern cross-package outcomes. This keeps the control proportionate while preventing scarce capacity from remaining invisible.

A fictional commissioning specialist is reserved for the current test window. The supply forecast improves and the network now permits earlier tests. The schedule display moves the test bar forward, but the specialist's reservation remains at the old window. The manager identifies a resource-feasibility gap: the earlier conditional finish is mathematically valid under the new durations, yet its start availability is unsupported. The app has no reservation field and cannot discover this gap itself.

The manager checks whether the reservation can be changed. Availability may be constrained by another project, qualification requirements or access arrangements. A substitute might be possible, but its competence and authority must be evidenced. The resource record distinguishes a named person from the capability the test actually requires. This prevents the team from treating any available technician as a valid replacement or assuming that the original specialist is indispensable without checking the rules.

The earlier test window also affects preparation. Procedures and instruments may have been planned against the old date. Bringing the test forward can create new readiness demand even when the physical system is complete. The manager reviews these obligations before issuing a revised commitment. The recovery action may need an enabling task or a different scope. A local supply improvement is real, but the integrated benefit remains conditional until testing capacity and preparation align.

If the earlier resource window cannot be secured, the sponsor can retain the original test date or choose another executable alternative. The network's unconstrained result remains a useful sensitivity, labelled as such. The operational forecast uses the actual reservation basis. These two results should not be blended into an unexplained date. The report should show why the theoretical saving is not currently available and what evidence could make it executable.

The decision also considers opportunity cost. Moving a specialist from another obligation can create exposure elsewhere. That consequence belongs to the organisation's resource decision, beyond this project's local network. A recovery plan that assumes resources can be reassigned without effect may simply transfer delay. The authority should state which obligation has priority and what residual exposure it accepts. The schedule cannot decide that organisational tradeoff from criticality alone.

The lesson is to review reservations whenever the forecast moves materially. Earlier completion can lose feasibility just as later completion can. Resource control should follow the current executable window rather than remain attached to a superseded baseline. The case supplies no quantified capacity or mobilisation rate. It exposes the distinction between a network-permitted start and a start supported by qualified, available capability.

Imagine the fictional installation uses a shared inspection service for equipment acceptance and receiving works verification. Both branches are expected to finish near the assembly release. The precedence network allows their inspections to overlap. The service has committed capacity for only one inspection at a time. This is a qualitative case, not a further frozen numerical study. Its purpose is to show an exclusion in the app: the network's early dates do not resolve how the shared service will be allocated.

The integration manager first checks whether the inspections truly need the same capability. If one can be performed by a different qualified authority, the conflict may be removable. If not, the manager compares readiness and consequence. Booking an inspection before the package is ready wastes capacity and can make the competing package wait. The resource decision therefore depends on credible readiness evidence as well as network criticality. The supplier and receiving owner should provide the evidence needed for a firm booking, with a rule for releasing unused capacity.

The manager next tests the interaction with acceptance. A preliminary inspection may identify defects without releasing assembly. A final inspection may require completed documentation that another team owns. Allocating the inspector does not solve a missing document dependency. The resource register and typed edge register should be reviewed together. Otherwise the team can spend effort accelerating an inspection whose output still cannot be accepted. The correct intervention may be earlier evidence preparation, a partial inspection with an explicit condition, or a later but executable booking.

If the manager adopts a sequence, the revised operational schedule should preserve why it was chosen. The underlying physical branches may remain parallel while the inspection tasks become resource-sequenced. When additional capacity later becomes available, the manager can identify exactly which relation to reconsider. This traceability prevents a temporary resource workaround from becoming a permanent unexplained constraint in subsequent baselines. It also supports a fair review of delay, since the record distinguishes package readiness failure from the consequences of an authorised shared-resource allocation.

Finally, the sponsor should know whether the resource arrangement is merely planned or actually secured. A forecast based on unconfirmed inspection capacity should be labelled conditional. The difference matters when downstream teams mobilise or external promises are made. Explicit conditionality can be uncomfortable, but it is more useful than treating a tentative booking as a completed readiness obligation. A resource plan becomes credible through evidence of availability and acceptance capability, not through the presence of a name beside each task.